Asset Details
Do you wish to reserve the book?
Convection‐Permitting Climate Models Can Support Observations to Generate Rainfall Return Levels
Hey, we have placed the reservation for you!
By the way, why not check out events that you can attend while you pick your title.
Oops! Something went wrong.
Looks like we were not able to place the reservation. Kindly try again later.
Are you sure you want to remove the book from the shelf?
Convection‐Permitting Climate Models Can Support Observations to Generate Rainfall Return Levels
Do you wish to request the book?
Convection‐Permitting Climate Models Can Support Observations to Generate Rainfall Return Levels
Please be aware that the book you have requested cannot be checked out. If you would like to checkout this book, you can reserve another copy
We have requested the book for you!
Your request is successful and it will be processed during the Library working hours. Please check the status of your request in My Requests.
Oops! Something went wrong.
Looks like we were not able to place your request. Kindly try again later.
Journal Article
Convection‐Permitting Climate Models Can Support Observations to Generate Rainfall Return Levels
2024
Overview
Information about the frequency and intensity of extreme precipitation is generally derived from fitting extreme value models using point‐observations, but the regionalization of these models is challenging. Here we propose using high‐resolution convection‐permitting climate model output as covariates for the estimation of observation‐based spatial rainfall return levels. We apply the Weather and Forecasting Research (WRF) model at a 1.5 km resolution driven by ERA5 reanalysis data over southern Germany, where 1,132 rain gauges provide observations of daily rainfall. For this complex topography, we build three different smooth spatial Generalized Extreme Value (GEV) models: (a) a reference model using latitude, longitude and elevation as covariates; (b) a model adding mean annual precipitation from the WRF; (c) a model adding extreme value statistical model estimates using WRF output. We show that the additional information provided by the WRF model can improve the representation of 10‐year and 100‐year return levels of daily rainfall by lowering the percentage bias, mean absolute error, and root‐mean‐square error. Furthermore, we conduct an extensive cross‐validation, where only 5%, 10%, 20%, 50%, 80%, 90%, and 95% of all rain gauges are considered when building spatial GEV models. Again, the additional information provided by the WRF model can improve results here. This cross‐validation study also highlights the robustness of our approach, showing great potential for use in data‐scarce regions. Plain Language Summary Heavy rainfall can trigger floods or landslides. In order to estimate the occurrence probability of these events, statistical models can be built using rainfall observations. However, these measurements are mostly point measurements and hence, one needs to interpolate in space when performing regional analyses. Often, topographical features such as elevation, latitude, and longitude are chosen as auxiliary variables to facilitate this interpolation. Here, we propose to add high‐resolution climate simulations as covariates. We compare three different setups which use data over southern Germany, where a dense rain gauge coverage is available: (a) a reference model using latitude, longitude, and elevation as covariates; (b) a model adding mean annual precipitation from the climate simulation; (c) a model adding extreme value statistical model estimates using data from the climate simulation. We show that the additional information provided by the climate model can improve the spatial representation of extreme daily rainfall. In most parts of the world, the rain gauge density is less than in the study area. Therefore, we test our three approaches with smaller subsets of the observational data. Our approach shows robustness under these conditions, highlighting potential for regions where observational coverage is scarcer but high‐resolution climate simulations are available. Key Points We utilize high‐resolution climate model data as covariates in spatial Generalized Extreme Value models to assess extreme precipitation Compared to using only topographical information, adding climate model data can improve the representation of 10‐ and 100‐year return levels An extensive cross‐validation study shows great potential of our approach for data‐scarce regions
